77A-AlOO 966 SCHOOL OF AEROSPACE MEDICINE BROOKS AFB TX F/ 5/10EVALUATION OF A MANIKIN PSYCHOMOTOR TASK. (U)MAY 81 D C READER, R A BENEL, A J RAHE

UNCLASSIFIED SAM-TR-81-10 NL

I IEEIIII[]IElllhEEllEEEEE

Report SAM.TR4I 10

C' EVALUATION OF A MANIKIN PSYCHOMOTOR TASK

David C. Reader, Wing Commander, RAFRucel A. Benel, Ph.D.Alton J. Rahe, M.S.

DT[*CSWEL E t

May 1901

Final Report for Period January 1980 - October 1980

Appwoved for public releas.; distribution unlimited.

USAF SCHOOL OF AEROSPACE MEDICINEJAAeospace Medical Division (AFSC)

rooks Air Force Baso, Texas 78235

7 07

NOTICES

This final report was submitted by personnel of the Crew PerformanceBranch, Crew Technology Division, and Data Sciences Division, USAF School ofAerospace Medicine, Aerospace Medical Division, AFSC, Brooks Air Force Base,Texas, under job order 7930-10-37.

When U.S. Government drawings, specifications, or other data are used forany purpose other than a definitely related Government procurement operation,the Government thereby incurs no responsibility nor any obligation whatsoever;and the fact that the Government may have formulated, furnished, or in any waysuppl ied the said drawings, specifications, or other data is not to be re-garded by implication or otherwise, as in any manner licensing the holder orany other person or corporation, or conveying any rights or permission to man-ufacture, use, or sell any patented invention that may in any way be relatedthereto.

The voluntary informed consent of the subjects used in this research wascobtained in accordance with APR 169-3.

This report has been reviewed by the Office of Public Affairs (PA) and isreleasable to the National Technical Information Service (NTIS). At NTIS, itwill be available to the general public, including foreign nations.

This technical report has been reviewed and is approved for publication.

DAVID C. READER, Wing Commander, RAF 0.IM9RPhDProject Scientist Supervisor

ROY L. DENARTColonel , USAF, 14CCommander

UNCLASSIFIlEDSECURITN CLASSIFICATION OF T.-IS PACE (Whe.,, 1I.-

REPORT DOCUMENTATION PAGE ________WEI___________IN(,_______

SAM-TR-81-1-0 *ft/1J6 5h

EVALUATION OF A MANIKIN PSYCHOMOTOR ;IOt

David C.'Reader Wg Cdr, RAF

Alto a A N .Ais*

USAF School of Aerospace Medicine (VNE)Aerospace Medical Division (AFSC) 62202F//Brooks Air Force Base. Texas 78235 1ZOP D79M TI-3

II CONTROLLING OFFICE NAME AND ADDRESS -FP

USAF School of Aerospace Medicine (VNE) ma M 81Aerospace Medical Division (AFSC) IT~ P OF PAOF0 ABrooks Air Force Base. Texas 7823.5 2314 MON ITORIN G AGEN CY N AME & ADDR FS5, It d-If-,,r I,-p, C-p11 *lh , I j~PIT. -7%

"Unclassified

16 DISTRIBUTION STATEMENT ,I tiA, Rep

Approved for public release; distribution unlimited.

17. DIST RiP auION ST A TEMEN T ,,t the nhstrn~lr: I pe- 11 l By 1 " k ,IrI rp

18 SUPPLEMENTARY NOTES

19. KEY WORDS (C-tinn- *.s-r- sIde I-1 .11 t-If ~thI A. SiA -- h

Manikin TaskPsychomotor Tasks and PerformancePlateau Performance, Objective and Subjective Performance MeasuresTask Differential Stability

20 ABSTRACT (C-6 ,~e -, ,,~, Id. If --- .. d IdenIIfv hv Ah-k -h-

A small, self-contained microprocessor displays a manikin on a TV screen invarious orientations. Subjects are required to choose lateral ities dependinqon a shape also presented on the display, and performance (accuracy and reac-tion time) is automatically recorded.

This task has been evaluated at the USAF School of Aerospace Medicine. Dataare presented to show that the task is simple to use, can be learned by allsubjects. and that Dlateau Performance can be auickly established.,-

DD I JA N73 1473 EDITION OF I NOV AS IS OBSOLETE UNCA5S''~SECURITY CL ASSIFIC ATION (S F 1-1% PAGE aN., P''. .,..d'

- A

UNCLASSIFIEDSECURITY CLASSIFICATION OF THIS PAGE( hen Del. E.teed)

20. ABSTRACT (Continued)

he various presentations were examined for difficulty; subjective and objectivemeasures of performance were compared; and repetitive effects were sought withsimilarities of presentation in sequence.

The analyses showed that control performance with this task was essentiallyconstant over the experimental period.

R A I

. I . TAB K

I _

iNIASS III P.0 , Itl! A %L& SIIC AT 0 OF J 1 1.1mI~,I ,,*,

CONTE NTS

Page

INTRODUCTION......................................................... 3

Format of Task .................................................. 4Experimental Protocol............................................ 5Experimental Procedure........................................... 5

RESULTS ............................................................. 9

DISCUSSION .......................................................... 19

CONCLUSIONS ......................................................... 22

ACKNOWLEDGMENTS......... ............................................ 23

REFERENCES .......................................................... 23

FIGURES

Figure

1. The manikin apparatus............................................ 6

2. A manikin presentation........................................... 7

3. Workload estimate ............................................... 8

4. Performance estimate............................................. 8

5. Task differential stability ..................................... 20

TAB LES

Table

1. Number of sessions (occupational level) ........................... 10

2. Number of sessions (frequency of training)......................... 11

3. Number of sessions (age) ........................................ 12

4. Sources of variation for the repeated measurements analysis ......... 13

5. Performance in the three groups (training frequency, occupationallevel, and age)............................................... 14

6. Performance and presentation type................................ 15

1

CONTENTS (Continued)

Page

7. Correlation between objective and subjective performance ratings... 17

8. Subjective vs. objective assessments ............................... 17

9. Mean reaction time for various presentations ....................... 18

10. Mean reaction time with varying degrees of similarity in sequence.. 18

11. Correlations of mean performance with time ......................... 19

2

EVALUATION OF A MANIKIN PSYCHOMOTOR TASK

INTRODUCT ION

Reaction time as a measure of human performance has been in use for over100 years. Response to a single stimulus is referred to as simple reactiontime, while the multiple stimulus situation is known as choice reaction time.Preference for reaction time as a measure of human performance has been predi-cated on the proposition that behavior takes time and the length of time re-corded as reaction time is an accurate appraisal of at least some segment ofhuman behavior. Since reaction time is commonly defined as the time from theonset of a stimulus to the initiation of the subject's response, cognitiverather than motor, processes are reflected in reaction time. To exploit thismeasurement fully, more complex stimulus arrays are necessary. For as WilliamJames (3) noted, it is only when the experiments are complicated that there isa chance for anything like an intellectual operation to occur.

Successful performance in modern military systems is, to a large extent,less dependent on skillful physical manipulation of controls, than it is oncareful planning of resource management and supervision of automated or semi-automated system operation. At the same time highly trained and skilled oper-ators are typically faced with complex stimulus arrays which nonethelessrequire discrete control actions. An experimenter who seeks to tap similarpsychomotor behavior in the laboratory with unskilled or minimally trainedsubjects faces a difficult decision in task selection. For these reasons morecognitively based performance measures that are not system dependent havebecome increasingly important for laboratory performance measurement. Themanikin task was developed at the Royal Air Force (RAF) Institute of AviationMedicine, Farnborough, United Kingdom, as a complex reaction time task thatrelated to performance in modern systems without being identified with anyparticular system.

For the manikin task (or any task) to be useful as a laboratory analog ofsome cor,,ponent of real-world system performance, it must meet several practi-cal criteria. Since time is generally limited for training of subjects,acquisition of asymptotic performance should not require protracted training.Also, it is impractical for all subjects to receive training on precisely thesame schedule. Therefore, asymptotic performance should not be training-schedule dependent. If preasymptotic performance is to be useful as ameasure, then the relative positions of individuals' performance should beconstant across some period of training and differential task stability (reli-ability) should be evident. For the task to have validity, it should be shownto be sensitive to manipulations of interest; e.g., the manikin test has beenshown to be sensitive to mild hypoxia.

Although the validity of the manikin task appeared well-established andhad been used to measure pciformance during hypoxia (2), reliability had notbeen systematically investigated. Additional factors influencing acquisition

3

of asymptotic performance also remained unidentified. For these reasons datacollection was instituted to verify the utility of the manikin task. Addi-tional analyses were undertaken to determine intratask variables that relatedto subject performance. Subjective impressions of performance were comparedwith actual measures of performance in a later phase of the experiment. Thesedata were also analyzed to seek repetition effects.

Format of Task

The apparatus consists of a microprocessor, power supply box, TV monitor,and two hand-held buttons (Fig. 1). A digital recorder is also required.

The manikin task presents a video picture of a little man (manikin) inoutline (Fig. 2). The arms are outstretched and hold a square in the righthand and a disc in the left hand, or vice versa. The manikin may be eitherupright or inverted and, in addition, may face towards or away from the sub-ject (four types of presentations). Facial and clothing details are visiblein the facing presentation. Below the manikin either a disc or square isshown. The subjects' task is to determine in which hand the manikin holds theidentical shape to that presented below it and, having decided, to press thebutton corresponding to that hand.

The manikin picture is generated in a small microcomputer which also con-trols the presentation timing. Each manikin is presented for 2 seconds andoccluded for 1 second. In addition, the manikin is occluded following eachresponse. A series of 96 manikins is presented in sequence. The sequence isfixed, but it always starts at a different place. The 96 presentations con-sist of 6 segments of 16 presentations, and each segment contains all the 16possibilities of presentation but in a different random order. Thus, for thepurpose of analysis, each segment of 16 (48 seconds) is of equal difficulty.The 96 manikin presentations make a period of 4.8 minutes. Four periodsinterspersed with 3 rest periods of 2 minutes are grouped to make a session of25.2 minutes. The experimenter presses the "prime" button on the microcom-puter to start a session of testing. Before each period the subject is warnedby a tone for 4 seconds before the first manikin presentation of that period.At the end of the last period, a warbling tone informs that the session isended and that the microprocessor timer requires restarting by pressing the"reset" button. Pressing the reset button before the end of the sessionresets the timer, and the prime button restarts it.

The microcomputer also generates a recording signal of the psychomotordata. When linked with a digital recorder, the following data are recordedafter each presentation: a number from 0-16 denoting the type of presenta-tion; after a space, a 1, 0, or * denoting whether a correct, incorrect, or noresponse was recorded; and after a further space, the reaction time in secondsto 2 decimal places. At the end of the session, a summary statement of totalnumbers of correct, incorrect, and no responses and the aggregate of reactiontime is recorded.

The logic of the microprocessor is fixed; neither the timing nor presen-tation sequence can be altered.

4

The manikin task data were recorded on the disc storage of a POP 11/34computer. An ASR 40 line printer was used in parallel to the computer torecord a hard copy of the data. A small switch box was employed whiereby indi-ces containing subject and run details could be inserted into the disc storageby the line printer.

Experimental Protocol

Two experiments were performed using the manikin task. The first soughtto examine the variables that govern acquisition of plateau performance, i.e.,those that control the shape of the learning curve. Eighteen naive subjectswere exposed to 10 sessions of training. They were divided into the followinggroupings: (a) by training frequency--6 subjects trained twice a day, 6 sub-jects trained once a day, and 6 subjects trained every alternate working day;(b) by age--below 28 years (7), between 28 and 38 years (7), and over 38 years(4); and (c) by occupational level--viz, professional scientists (6), noncom-missioned officers (6), and airmen (6). Subjects were not permitted to exam-ine the data during the study.

The second experiment examined the relationship of subjective and objec-tive measures of performance. Subjective assessment of performance has beenused extensively where direct objective measures of performance are denied.The same 18 subjects were exposed to five further sessions with the manikintask, one per day, on consecutive days. They were asked to estimate both theworkload effort and their performance against seven-point scales (Figs. 3 and4). This experiment also examined whether the shape (disc or square) or side(right or left) affected overall performance and also checked the difficultyof each of the four types of presentations for the subjects.

Repetition effects were also considered using data from the second exper-iment. The manikin task sequence is fixed; thus all subjects were presentedwith identical sequences (but different start positions). The similarity ofpresentations to the immediate preceding presentation varies as shapes, sides,upright or inverted, forward and away facing, are altered. The sequences per-mitted 7 levels of similarity to be compared from exactly comparable to grossdissimilarity (every variable reversed). The data from the sessions conductedduring the second experiment (subjective versus objective assessment) wereanalyzed to detect repetition effects, i.e., enhanced performance with simi-larity of presentations in sequence. Finally, an analysis was performed onthe stability or reliability of the task data over the entire period of theexperiment.

Experimental Procedure

Subjects sat in a darkened, quiet room, viewed a 7-inch (17.78 cm) videodisplay from approximately 38 inches (96.52 cm) (approximately 10' visualangle) and held press-button boxes in either hand. Only one session of per-formance (25.2 min) was conducted for each subject at a time. In some i n-stances, ear defenders were used when external noise could not be controlled.An experimental monitor briefed the subjects and controlled the apparatus andcomputer. Distraction to the subjects was minimized at all times.

5

- 7-7 -1-

INSTRUCTIONS- -,r'cle the number that best describes the workload you experi-enced while performing the task during this session.

1. NOTHING TO DO; NO SYSTEM DEMANDS

2. LITTLE TO DO; MINIMUM SYSTEM DEMANDS

3. ACTIVE INVOLVEMENT REQUIRED: EASY TO KEEP UP

4. CHALLENGING, BUT MANAGEABLE

5. EXTREMELY BUSY; BARELY ABLE TO KEEP UP

6. OVERLOADED; HIGH CHANCE OF ERROR

7. UNMANAGEABLE; UNACCEPTABLE

Figure 3. Workload estimate.

INSTRUCTIONS: Circle the number of the appropriate term. Compared to my per-formance last session, my performance on this session was ............

1. MUCH WORSE

2. WORSE

3. SLIGHTLY WORSE

4. SAME

5. SLIGHTLY BETTER

6. BETTER

7. MUCH BETTER

Figure 4. Performance estimate.

The data on disc storage was transferred to an IBM 360 computer where allanalyses were performed using the Statistical Analysis System (SAS)

RESULTS

Experiment I examined the acquisition of plateau performance. For thepurpose of the experiment, plateau performance was defined as not exceeding± 5% of the mean reaction time (RT) of the previous 2 sessions. The numaber ofsessions required to achieve this criterion are given in Table 1. The dataare classified by occupational level. In one case, a strategy adopted by onesubject (JB) allowed him to improve his performance throughout the experi-ment. Therefore, he did not reach the criterion set and was excluded from theanalysis. Table 2 classifies the data by frequency of training. Table 3classifies the number of sessions to plateau by age. Both a one-way analysisof variance and a nonparametric H test on ranked data failed to detect anysignificant differences in number of sessions to plateau between any of theabove three groupings (training, occupational level, and age) at the 0.1level.

A repeated measurements analysis of variance procedure was used to ana-lyze the reaction time (RT) data for training group, session, and period dif-ferences, plus their interactions. (See Table 4 for the sources of variationin the analysis.) There was strong evidence (p<.001) that the pattern of "heperiod means was not consistent from one session to another (period by sessioninteraction). This inconsistency is a reflection of more improvement overperiods for the initial sessions than for the latter sessions, a finding whichwas expected. The analysis failed to detect any differences among the threetraining groups in this improvement in performance. That is, the traininggroup did not show any significant interaction with period or session. Kfcourse, the overall performance improved over the ten sessions (p< .001).

There was some indication of an overall difference arong the three train-ing groups (p<.10). (See Table 5 for means.) The group that trained onalternate days had a lower mean reaction time than that of the other twgroups. The repeated measurements analysis procedure was also used to -, yzethe "number of correct responses" and the mean reaction time for the c rectresponses. The testing results were very comparable among the analyses ferthe three measurements. Again the group that trained on alternate days had abetter accuracy mean (more correct responses) and lower mean reaction 7e forthe correct responses only. It must be remembered that thesy are .,ralltraining group mean differences (averaged over all sessions) and the are donot reflect a training effect.

The associated mean reaction time, etc., are also given in Table 5 forthe age and occupational level groupings. Generally, no significant differ-ences were detected among these groupings. The analyses on each of the threegroupings (training, occupational level, and age) were performed ignoring theother two groupings.

Further analyses of the data showed that the manikin presentations Imsedvarying degrees of difficulty. Table 6 lists the mean reaction time for thethree groups divided by presentation type.

TABLE 1. NUMBER OF SESSIONS TO CRITERION (PLATEAU PERFORMANCE)(CLASSIFIED BY OCCUPATIONAL LEVEL)

Group Subject Number Mean Min Max

Professional WS 4scientist SG 4

DB 10LP 5PD 8DM 5

6 4 10

NCOs BW 5PP 7EC 7FA 5DT 5JB * (Did not reach plateau)

5.8 5 7

Airmen RC 6TB 7VP 9TW 5JS 5MF 7

6.5 5 9

10

TABLE 2. NUMBER OF SESSIONS TO CRITERION (PLATEAU PERFOROANCE)(CLASSIFIED BY FREQUENCY OF TRAINING)

Group Subject Number Mean Min Max

Twice daily PD 8DM 5FA 5OT 5JS 5MF 7

5.8 5 8

Once daily DB 10LP 5PP 7EC 7VP 9TW 5

7.1 5 10

Alternate WS 4days SG 4

BW 5JB * (Did not reach plateau)RC 6TB 7

5.2 4 7

ml

TABLE 3. NUMBER OF SESSIONS TO CRITERION (PLATEAU PERFORMANCE)(CLASSIFIED BY AGE)

Group Subject Number Mean Min Max

>38 WS 4DB 10PD 8PP 7

7.25 4 10

28-38 SG 4LP 5DM 5FA 5EC 7BW 5JS 5

5.1 4 7

<28 JB * (Did not reach plateau)DT 5RC 6TB 7VP 9TW 5MF 7

6.5 5 9

12

TABLE 4. SOURCES OF VARIATION FOR THE REPEATED MEASUREMENTS ANALYSIS

Degrees ofSource freedom

Between training groups 2Between subjects* within training groups 14

Between sessions 9Interaction of training groups and sessions 18Interaction of subjects and sessions within groups 134

Between periods 3Interaction of training groups and periods 6Interaction of subjects and periods within groups 44

Interaction of sessions and periods 27Interaction of training groups and sessions and periods 54Interaction of subjects and sessions and periods within

groups 404

*Data for JB were not included in the analysis.

13

TABLE 5. PERFORMANCE IN THE THREE GROUPS

Mean RT (sec) for

Group ERT (sec) No. correct correct responses

Training frequency

Twice daily 78.18 90.8 .809Once daily 88.22 91.4 .912Alternate days 68.24 93.1 .709

Occupational level

Professionalscientist 83.38 91.5 .865

NCOs 79.37 91.1 .820Airmen 71.08 92.7 .738

Age

<28 70.57 92.6 .73228-38 79.51 92.1 .822>38 87.15 89.9 .3,1

14

TABLE 6. PERFORMANCE AND PRESENTATION TYPE

Group Mean reaction time (sec) for correct responses

Erect Erect Inverted Invertedaway facing facing away

Training frequency

Twice daily .741 .803 .830 .864Once daily .845 .921 .929 .956Alternate days .643 .728 .705 .762

Total .793 .817 .821 .860

Occupational level

Professionalscientist .799 .878 .869 .907

NCOs .729 .798 .866 .896Airmen .694 .769 .720 .772

<28 .684 .750 .727 .77128-38 .740 .815 .846 .wq?>38 .826 .913 .9?3 .q47

A repeated measurements analysis was performed for each presentation typeseparately using only the RT values for which the associated responses werecorrect in order to eliminate the reaction times with the incorrect re-sponses. The results for each analysis were very comparable to the previousresults; i.e., no additional meaningful information was gained. Generally,the erect away presentation had the lowest mean reaction time with the erectfacing, inverted facing, and inverted away groups having progressively in-creasing reaction times. The means are also given in Table 6 for the occupa-tional level and age groupings. No statistical tests were performed on thesegroupings.

The second experiment examined the correlation of subjective and objec-tive measures of performance. After each session the subjects reported on ascale from 1-7 their impressions of workload and performance. These data werecompared with the performance data (RT) from the manikin task. As subject JBused a different strategy for the task than all the other subjects, his datawere omitted from the analysis. Table 7 lists the Pearson product-moment cor-relation coefficients between the mean RT data (for all sessions) and each ofthese two subjective values among the 17 subjects. These values are low anddo not differ statistically from zero. Table 8 lists the correlation coeffi-cient for performance and RT as well as workload and RT among the 5 sessionswithin each subject separately. Only one (.898) of the coefficients differedsignificantly from zero. However, each coefficient is based on only 5 pairsof data. The pooled correlation coefficients do not differ significantly from

zrThe reaction time data for the second study were also used to test forshape (square vs. circle), side (right vs. left), and type of presentation (4)differences. A repeated measurement (shape, side, type, and session) analysisof variance procedure was used to test for differences in reaction time. Ses-sion did not show any significant interaction with any of the other threefactors. Table 9 lists the mean reaction time for the various types of mani-kin presentations. The pattern of the RT means for shape and side are gener-ally quite comparable for the different presentation levels. The types ofpresentation were ranked erect away, inverted facing, erect facing, and in-verted away in ascending order of reaction time and difficulty (p=.001).Table 9 also shows the effect of shape and side. Discs did not differ fromsquares, but there was a significant (p<.001) difference of right from left.The right-sided responses were approximately 25 msec faster than the left.

Performance with emphasis on the similarity of successive presentationswas also studied. All presentations were classified by degree of similarityfrom identical (1) to gross dissimilarity (7) as described in Table 10. Themean reaction times for the second presentations of each successive pair forthese sequences are listed in Table 10. The order of similarity is not echoedexactly in the mean reaction time ranking, but a similar trend is seen. Thereaction times with the grossly dissimilar sequences (similarity 6 and 7) dif-fer significantly (p<.OX) from the moderately dissimilar sequence reactiontimes. In order to equalize the variances in this analysis, a log transformof each RT value was used in a two-way (proportional) analysis of variance.

16

TABLE 7. CORRELATION BETWEEN OBJECTIVE AND SUBJECTIVEPERFORMANCE RATINGS

RT vs. workload RT vs. performance

Mean RT across all sessions (N=17) .028 -.337

TABLE 8. SUBJECTIVE VS. OBJECTIVE ASSESSMENTS

Reaction time compared withSubjects Performance Workload Performance Workload

Min Max Mi Max

WS .048 -.714 4 5 3 4SG -.595 .786 3 6 2 4DB -.838 .726 3 5 1 2LP -.189 .758 3 6 2 3PD -.100 .000 3 6 4 4DM -.488 -.253 3 6 2 3

BW -.075 .000 4 5 2 2PP -.286 .323 3 5 2 3EC .000 .000 4 4 2 2FA -.434 .463 3 5 2 3DT -.178 .000 3 5 2 2JB Excluded

RC -.594 .787 3 5 3 4TB -.032 .898 2 7 2 4VP .311 -.809 3 5 2 3TW -.189 .492 1 4 2 3JS -.221 .000 3 7 3 3MF .270 .000 4 6 2 2

Pooled -. 244 .264

17

TABLE 9. MEAN REACTION TIME FOR VARIOUS PRESENTATIONS

Presentation

Erect Erect Inverted InvertedShape Side away facing facing away Total

Circle Left .626 .664 .670 .713 .668Right .610 .651 .637 .678 .644

Square Left .630 .670 .658 .705 .666Right .612 .635 .648 .664 .640

Total .620 .655 .653 .690

TABLE 10. MEAN REACTION TIME WITH VARYING DEGREES OF SIMILARITY IN SEQUENCE

Similarity Description N* Mean RT Order

1 Same presentation, shape, side 19 .615 1

2 Same presentation, side 39 .636 5

3 Same presentation 156 .619 4

4 All erect, same side 373 .618 3All inverted, same side

5 All erect 314 .616 2All inverted

6 All different but same side 471 .644 6

7 All different 510 .645 7

3, 4, 5 differ significantly from 6, 7. (p<.O1)

*Number of tasks/subjects.

18

A major criterion of any psychomotor task is that it should produce sta-ble results in stable conditions; i.e., the differential stability should behigh. Bittner (1) has suggested that all tasks should be examined for stabil-ity and proposed a graphic analysis strategy for this. Pearson product-momentcorrelation coefficients were calculated between all pair-wise session means(1-15) using the reaction data with only correct responses for the 17 sub-jects. These are listed in Table 11. The lowest between-sessions correlationwas .56 (sessions 1-13), indicating a fair degree of stability even withoutextensive practice. A selection of these correlations is plotted on Figure 5for sessions 1, 2, 4, 6, 10, and 12. Moreover, the manikin task may be seento have good task reliability with an estimate of the common correlation of.84.

TABLE 11. CORRELATIONS OF MEAN PERFORMANCE WITH TIME

Session 2 3 4 5 6 7 8 9 10 11 12 13 14 15

1 .82 .73 .70 .77 .71 .68 .60 .74 .65 .62 .59 .56 .65 .662 .94 .94 .95 .89 .88 .80 .87 .90 .77 .77 .73 .77 .823 .98 .98 .94 .89 .86 .87 .92 .69 .72 .67 .72 .854 .98 .95 .91 .87 .88 .93 .73 .77 .70 .75 .845 .95 .91 .87 .92 .92 .72 .76 .71 .74 .876 .94 .91 .91 .94 .71 .75 .69 .75 .827 .97 .97 .97 .85 .85 .78 .84 .868 .95 .95 .80 .81 .76 .80 .879 .93 .80 .81 .74 .79 .85

10 .84 .87 .82 .85 .8911 .97 .92 .94 .8112 .93 .92 .8413 .95 .901415

DISCUSSION

The choice of a criterion for plateau performance was based on acceptabletolerance levels for control sessions for individual subjects. That all tNOone subject achieved plateau performance in the confines of the experimentalexposures was gratifying. The strategy adopted by one subject (JF) was inter-

esting and unique. fie reasoned that the logic of the manikin microlrrocessordemanded a fixed sequence, albeit with a random start position. hav inq io-cided that, he sought to val idate his hypothesis by memorizing sequeWnces anodhunting those sequences in later exposures. In that event he was very s'c-cessful, lie demonstrated memory scans of Lii) to 30 sequences where the reac-tion time was zero and correct responses were recorded. Accordingly, as hisremory bank grew, he achieved better performance and never acheved a ;]Ia-teau . As he was not dissuaded from that strategy, he ix*rsisted; hm, his dataare misleading and were excluded fromn the analysis.

,.1I

V\

~~C14

*0 0 0 0 .C

SJN 11A30 NIilHO

T J'720

No subject achieved plateau performance with less than 4 su,' Ios oftraining, but all had reached the criterion in 10. Fhe a'ean was .jP, ardthis figure gives good guidance for future training sessions. The datat failedto suggest that frequency of training, age, or occupational level affect therate of achievement of plateau performance which enhunces the ti 1 it y of thetask. Admittedly, the age span used in the experiment was siail , bht it wasthe ~maximum practical, given the available subjects.

Although the number of training sessions required for plateau Performianceshowed no appreciable differences with the groups used, the overall p)erfor:-ance did provide so;me suggestions that the group that trained on Il ternatedays had both a lower ;aan reaction tiine and a better record of accuracy.However, the number of subjects in the training groups was smiall (6) and ,heseobserved, albeit significant, differences were due vlore to individi:,l differ-ences than oredictable group distinctions. Furthernore, the gr)ou' "s of aueand occupational level were not so distinguished, and these data were reco rdedbefore plateau performance had been achieved.

The analysis showed that all sub iects cons dared "he +.,e, ) , i , ,

presentation as the fol Iowin g order of difficuIty as graded by the mean re ac-tion tiless for e, ch: erect away, erect licing, inverted facing, and invertedaway. This corresponds to previous analyses which led to the adopt;,, Iou ', theimnikii software (,f the even distribution of all possibilities of presentationin ea.ch It,, resunta t ois. This permits analysis of iperforance for e,,ch 1,resentation segment, if need be; e.g., for rapid changes in performnc, )r

consciousness. However, no such segm-ental analyses were perfoied ,' thisstudy. The order of difficulty is a reflection of the comlplexity of tne t-tal reorientation required to imatch Lhe "'a riin to !hat of the sob ec . Theiright stance was clearly easier than the inverted, the facinl op .iv n

than facing towards. The order )' difficulty was prserved icro-tliency of training groups, with four imnor exce A ians

In the second expieriment a similar an rl VO ' ,0 ,; I'. a1 tnio this seqenoi - I .e ., the excha'ne i !,he ramk I erect ti an(,verted facin,, . _xperiment I data were from sh iec ts who were learnt AO0,,.,. I ho,, tro L.xperirtent I cae fron trained iersonnel '.he difference illthe nrd!,' tay indicate som e changes in strate.y with learning hj', is ;)rehabivri10t. w,,,, h close examinaition given the small ditferences in ,:ean r 4C + n i

1 t.he r a V I i q .

The second experim;ient was an atftmwe to \ lilate the sub'ect vo ra;iinocards used by the (crew Performnance i[ranch ,)f !he 'It ,ctho, of .ter's1" "'nd-icine. However, these cards are of ore utility when cnnditi t. are vario!e ,e.g., subjects becoming exhausted, workloads varying t ro'- i na(t iv 1y 'o over-1 odd , etc . In this experimlent the workload pre-onted was idontical each ti!ei,and perfortarice varied 1 itt Le. . oe subjects reft]en, ed l hii their 1,lvaril-

ate assess;!ients . Accordingly, little correlatittn ws exected he Iwee r hi en-ti ve and obiec .i v(l Issessments and I itt. Ie was ob!served. %1 :w siblec sJ 'wet1better correl at. ions than oth ers, s, ', ar t icl ,rl y wit h w t- 1 td, ') 1 he ,* ocorrelations were not signi ficantly distribuei to ay on' of he ,irt roi p.

Ihi s tart of the experiment was a minor tjoal of t he st ly and lies It, r o* t,t, he val i dIy if the concept of suhjPctive assesents ]sld e- ,i 1 r ein very ,table conditino, as in this ex;, r

Hunting for discs rather than squares in the task presented m) chd,,w, 1w,

difficulty as almost identical mean reaction times were obtained tur r'jrh.However, there was a considerable difference in the right-hand-l I '

compared to left. All but two subjects were right-handed, and the wnmference undoubtedly reflects this handedness bias.

The second experiment also examined the alteration of reaction ": bysimilar previous presentations. The memory of a previous mani ;rnb1solved should have made the solution of a similar problem easier ai 0d nce rc--duce the reaction time. These analyses confirmed this hypothesis (,f In,,tition effect.

The analysis showed that the logical similarity ranking dil ,, ' -

spond exactly to the reaction time ranking. Rank 2 which presen ,,i ,,presentation and the same side as the previous mani kin had miuch I , -! -tion times than those where side and shape were different (2 4.i ,.when all three variables were altered (types of presentation, side. . ,

the reaction times were 30 msec longer.

The results of the analysis of differential stability (7db,

Fig. 5) are encouraging. They show the repeatability of the tas .one session to another. Some instability due to learning is to b -note the consistently lower correlations obtained: with sessionlowest correlation observed is .56.

The plots in Figure 5 show no overall trend, up or down, h

changes occur between sessions 10 and 11. These changes were ciJ-!weekend intervening for all subjects between experiments 1 andstudy.

CONCLUSIONS

1. The manikin task is simple and reliable to Lise.

2. All subjects can learn the task and acquire pat eau performancfl.

3. Acquisition of plateau performance is achieved in appr , ,sions of practice; the extremes range from 4 to 10 sessions.

4. The rate of acqui si ti on of pl ateau performance is essent il ...

of the frequency of practice or occupational level or age ()f th ,

5. The various presentations do represent varying difficultie,but this variation is constant for all subjects. The sequence, of I,-1

ensures that all 16 possibilities (and difficulties) are posed every .

6. Right-handed responses were quicker than left-handed , e .

squares and discs produced similar responses.

7. There was poor correlation between suhjecti ve aind oh ec Ie1 ',

performance , but, this was due to the constant control crmd i t ,'i .

than the inadequacies of the performance nea sires.

A repet it ive( eftt occ urs wi th :v'il it~ , !r i t-)L'lence . However , the 1uini k i n talsk has ai t i ed m..'.f ec t is Coameron tor a] I suhi ec t s andI5~u 1 on s

0. The task has a high vraieot Ji f t ertI i .il I AS y, 1 1 ~ o r

per f or, ianrce i s esSent il1 y conta~Lnt w it h i irie.

The will ig and courteous caJO;era!0 ion of the st t t m ,e prf )r- Iaice Branch of the USAF School of Aeraspa'ce Medi cine is gratef..1ly 10C -)Iwl-t-due ld .Spec i,- rI mont i on must be nao f)r* ol ectron ic a-si stance (jiv en tiy 3Ve

A. shbory , and experi ;iental 1Tini taring by Sgjt Po in Chave(z , sr, , >r-

InIk S, an~ld ALI.C Va 1 ia )n Pidrnt!

I Biit tn ur , Ak. C.Jr S ta t is t i calI tst; for d if fe re nt ialI stadbl itv. 'r-ce ed i negs o f the Hum1ran F ac t. ,r; S)c ie ty 2?3rd Annual Meet 1 rig. , oston,Mass , 1979 .I

efuIcoriscioutnfess follIowi nq racpijd deIcogrPIess ion t ra-i 10() t,) ?2C,i)()

Or' 27 ,000 ft . 45 ~ri Syiipo si i Aeros pace Med ical Ass cci at ion, Wash -

i nqton , 1).C, 1973.

3. davies, W. Ps ycho]I ogy The 1tri efer (curse. Ne w Yok j o r 1k9